Update of modeling_cogvlm.py for Transformers newer version
#15
by
Qishuai
- opened
- modeling_cogvlm.py +15 -4
modeling_cogvlm.py
CHANGED
@@ -1,9 +1,11 @@
|
|
1 |
"""largely copy from llama and adapt for cogvlm"""
|
2 |
import warnings
|
|
|
3 |
from typing import TYPE_CHECKING, Optional, Tuple, List, Union, Literal, Dict, Any
|
4 |
|
5 |
import math
|
6 |
import torch
|
|
|
7 |
from torch import nn
|
8 |
from torch.nn import CrossEntropyLoss
|
9 |
from torchvision import transforms
|
@@ -26,7 +28,12 @@ logger = get_logger(__name__)
|
|
26 |
|
27 |
LANGUAGE_TOKEN_TYPE = 0
|
28 |
VISION_TOKEN_TYPE = 1
|
29 |
-
|
|
|
|
|
|
|
|
|
|
|
30 |
|
31 |
# Copied from transformers.models.bart.modeling_bart._make_causal_mask
|
32 |
def _make_causal_mask(
|
@@ -736,9 +743,13 @@ class CogVLMForCausalLM(CogVLMPreTrainedModel):
|
|
736 |
standardize_cache_format: bool = False,
|
737 |
) -> Dict[str, Any]:
|
738 |
# update past_key_values
|
739 |
-
|
740 |
-
|
741 |
-
|
|
|
|
|
|
|
|
|
742 |
if getattr(outputs, "state", None) is not None:
|
743 |
model_kwargs["state"] = outputs.state
|
744 |
|
|
|
1 |
"""largely copy from llama and adapt for cogvlm"""
|
2 |
import warnings
|
3 |
+
import packaging.version
|
4 |
from typing import TYPE_CHECKING, Optional, Tuple, List, Union, Literal, Dict, Any
|
5 |
|
6 |
import math
|
7 |
import torch
|
8 |
+
import transformers
|
9 |
from torch import nn
|
10 |
from torch.nn import CrossEntropyLoss
|
11 |
from torchvision import transforms
|
|
|
28 |
|
29 |
LANGUAGE_TOKEN_TYPE = 0
|
30 |
VISION_TOKEN_TYPE = 1
|
31 |
+
TRANSFORMERS_ABOVE_441 = (
|
32 |
+
True
|
33 |
+
if packaging.version.parse(transformers.__version__)
|
34 |
+
>= packaging.version.parse("4.42.0")
|
35 |
+
else False
|
36 |
+
)
|
37 |
|
38 |
# Copied from transformers.models.bart.modeling_bart._make_causal_mask
|
39 |
def _make_causal_mask(
|
|
|
743 |
standardize_cache_format: bool = False,
|
744 |
) -> Dict[str, Any]:
|
745 |
# update past_key_values
|
746 |
+
if TRANSFORMERS_ABOVE_441:
|
747 |
+
cache_name, cache = self._extract_past_from_model_output(outputs)
|
748 |
+
model_kwargs[cache_name] = cache
|
749 |
+
else:
|
750 |
+
model_kwargs["past_key_values"] = self._extract_past_from_model_output(
|
751 |
+
outputs, standardize_cache_format=standardize_cache_format
|
752 |
+
)
|
753 |
if getattr(outputs, "state", None) is not None:
|
754 |
model_kwargs["state"] = outputs.state
|
755 |
|